OpenAI Slows Development Track Amid Concerns ‘Progress’ Is Outpacing Safety

OpenAI says it is pausing AI testing and slowing research amid concerns that safety standards and procedures are not keeping pace with the development of its models. Last month, OpenAI said that some of its AI models collaborated to find vulnerabilities in the company’s systems. Although OpenAI employees discovered the message board and cleared the messages, the company said that its models began collaborating again to exploit vulnerabilities and access the internet. One of its models hacked a separate company, Hugging Face. After that hack was disclosed, other AI research companies, such as Meta AI, said that their models had hacked other companies. In a blog post on Tuesday, OpenAI wrote, “As models become more capable, the risks associated with developing and testing them internally also grow.”“Our standards for monitoring, alignment, and security must stay ahead of those risks. We wanted to take the time necessary to meet those standards, so we temporarily slowed the pace of scaling,” the company added.The company noted that its Astra model may have “critical cyber capabilities,” under its preparedness framework, which means that it can discover and exploit security vulnerabilities without human involvement.OpenAI also said it is pausing for two weeks its reinforcement learning training — where models learn without human involvement — and that it is working to bolster the security environments in which it tests its models.The CEO of OpenAI, Sam Altman, wrote in a post on X, “Model progress is now extremely rapid, and we always said we would take action if we felt that model capabilities were outstripping the pace of safety and alignment.”“We care very deeply about AI safety. We believe the entire field will have to coordinate on shared safety standards, but will act unilaterally in the meantime,” he added. The announcement comes roughly a week after Senator Bernie Sanders urged tech companies to pause AI development. In a letter to AI companies, the Vermont senator said, “Almost every day, there is a new story about how your companies are losing control of the AI technology you are developing, with potentially cataclysmic results.”“And yet, at a moment when we have seen human loss of control and the creation of potentially dangerous viruses, your companies are still racing ahead — investing tens of billions of dollars into a technology that nobody can fully understand, predict or control,” Mr. Sanders said as he threatened government action if the companies did not voluntarily limit their research.

OpenAI Slows Development Track Amid Concerns ‘Progress’ Is Outpacing Safety

OpenAI Slows Development Track Amid Concerns ‘Progress’ Is Outpacing Safety

More from the New York Sun

An Eminent Historian Turns Militant Over Israel

Trump’s BBC Lawsuit Reassigned To Judge Who Helped Keep His Lawsuit Against the Pulitzer Prize Board Alive

America Imposes New Sanctions on ICC Officials to Chorus of Globalist Boos

A Female Movie Pioneer Who Showed Britain ‘The Way We Live’

Aston Martin’s New Valen Is a Vulgar Compromise

AOC, a Poster Child for Economic Illiteracy, Is Unfit To Serve as President

Naked in the River of Time

Two Pro-Israel Jewish Democrats Easily Survive Primary Challenges from the Left in Florida

More from the New York Sun

An Eminent Historian Turns Militant Over Israel

Trump’s BBC Lawsuit Reassigned To Judge Who Helped Keep His Lawsuit Against the Pulitzer Prize Board Alive

America Imposes New Sanctions on ICC Officials to Chorus of Globalist Boos

A Female Movie Pioneer Who Showed Britain ‘The Way We Live’

Aston Martin’s New Valen Is a Vulgar Compromise

AOC, a Poster Child for Economic Illiteracy, Is Unfit To Serve as President

Naked in the River of Time

Two Pro-Israel Jewish Democrats Easily Survive Primary Challenges from the Left in Florida